PLan:
make a halfbridge with IGBT's, without body diodes, to be used with symmetric power supply.
switched coil, is grounded to earth ground, by alternatly switching positive or negative, I create positive and negative impulses.
At the same time, I also connect a series resonant coil to ground, which will provide the low impedance path to ground for the impulses, this series resonant coil will be the primary.

IGBT's will need proper cooling to stay stable. Altough, they should not get very hot.

L1 (impulse generator coil) and L2 (series resonant impulsed coil) are both pancake bifilar coils, and close coupled.
L3 wil be loose coupled to L2, L3 is the secondary, set into LMD resonance producing power.

I made this half bridge before, but with mosfets, which always have body diodes. This required the impulses to travel trough extra diodes, whic became a great loss of heat radiation. which destroyed efficiency.

Igbt's won't need these diodes, so the impulse can travel straight into L2. without heat loss, much more efficient.


this IGBT specs look good, although not as fast as a mosfet, I will need to give it a try.

RGS50TSX2DHRC11 also known as RGS50TSX2HR
1200V 25A 396W Vce 1,7V
turn off delay time: Td(off)=140nS
fall time Tf = 205nS

igbt mouser link

spec sheet

ordered a few, and see how that performs. can use the gate driver circuit I already have... nice.
gate emitter threshold 7V  maximum voltage 30V, so 20V will work fine

NoTe! both IGBT's need isolated gate drivers! I already have them, but will design new ones with isolated dc dc converters 12V in 20V out 2W

this curve should suggest to keep the gate to emittor voltages around 10 to 11 V instead of 20V

this keeps the charge low, which should make turn off faster.

Right?
just guessing
never really worked with IGBT before

that sounds very good. it compares directly to your previous circuit to see if this is better. Maybe you can run a simulation with LTSpiec before.

Body diode of the mosfet was a problem with the half bridge. But it also protected the mosfets from over voltage, as avalanche diodes.

With IGBT, without body diodes, there is no more protection against over voltage.

Normally TVS diodes could be used for protection, but now I have both polarity impulses inbetween the high and low IGBT's.

This makes it very vulnerable.

 so they can be used. several in series,
During tuning the impulse voltages change. So extra care must be used while tuning. Tuning with lower voltages, and only boosting the input power, when it's safe (stable).

IGBT's can handle 1200V so that's not easily reached. but still possible.

Maybe there is another way of protection possible, that works with both polarities...

edit
I remember TVS can be bipolar or mono polar, so placing several in series, in both directions should work.
one direction is open, while the series connected other direction is blocked.
this makes it work in both directions.
Pfew! that will save a lot of IGBT's

I ordered 1.5KE440CA  TVS diodes.
2 will be put in series, parallel over one of the IGBT's
these will protect the IGBT's from over voltage from both polarity impulses.

impulses can still rise up to around 900V

I wonder if If IGBT turn off can be quicker when gate is negative relative to emittor

awesome, IGBT's already have arrived. TVS diodes will come later today.
Much sooner than expected.
Post system seems to work again.

1EDI60I12AF gate driver

I keep loosing this one... its the gate driver IC

Quote from evostars on September 23rd, 2020, 06:57 AM

1EDI60I12AF gate driver

I keep loosing this one... its the gate driver IC

What do you mean? Was it destroyed by a voltage pulse?

Quote from jan_2020 on September 23rd, 2020, 11:01 AM

What do you mean? Was it destroyed by a voltage pulse?

no, I lost the pdf :)

I will again add a 4th coil. close coupled to L3
this time series resonant, with cap grounded.
I want to see If I can rectify and load this 4th coil.

but first I need to rebuild the half bridge, and test it with IGBT's and tvs diodes
see how it performs

You have already tried your attempt with the H bridges this year. What is different in the current attempt than 6 months ago?

Quote from jan_2020 on September 24th, 2020, 03:33 AM

You have already tried your attempt with the H bridges this year. What is different in the current attempt than 6 months ago?

H bridge is left behind because the impulse needs an earth ground.

H bridge is a double half bridge.

half bridge I tried with mosfets. but body diodes are a problem.

So now I use IGBT's without body diodes in a half bridge.

If that works, 2 IGBT half bridges could again become a H bridge (depending on its function).

I hope that the dead time of the igbt driver is long enough for your current setup. As I now know all your posts here and on YouTube you will find a solution.

Did first test with new IGBT half bridge.
series resonant at 67.13kC
with L2 series cap =61nF
power in is around 67W
resonant voltage 1.17kV
resonant amps 28.5A(!)
 :shocked:

but... I haven't seen a spike yet.
tried duty cycle at 45% still no spike.
suspect turn off of IGBT is slow.

but this is first test. might be something wrong. Although the sine wave does jump up at its max as before. so that indicates L1 dumps it magnetic field energy into L2

this is the (basic) circuit right now.

gate is driven to 20V to 0V
might be to high. could make it 13V and - 7V  with emitter at zero V (relative)

IGBT's are active cooled by fans on heatsinks.

If I switch one IGBT than impulse is visible.
1.3uS duration. slow but this can be improved.

I also tested without L1, just series resonant L2. Shows a great improvement of efficiency when L1 is part of the circuit.

very nice to see this all working.

Now I need to put L3 on again, make it parallel resonant, so I will have 2 L2 series resonant frequencies again (loose coupled) .
The highest L2 frequency is the LMD where by the impulse becomes faster as it is sucked into L3. this will also make the impulse higher in amplitude /voltage

I now switch the IGBT with 20V at the gate.
I made this from series connecting a charged 6V and 12V battery, both in series.

by removing the 6V battery, I will switch the gate with around 13V.

the gate switches fully on around 10V. So this 13 instead of 20V should build up less charge in the bjt part of the igbt.

Less charge to discharge should make it turn off faster.

this is an easy test to see the difference.
It should shorten the impulse duration.

compared 19.56V VS 12.96V at the gate, didn't do anything. will continue using 12.96V (12V lead acid).

plan is to use the 6V battery, to create -6.5V but I need a 2 new isolated gate driver boards for that.

will continue with loading L2 with Parallel resonant L3

added L3 and L4 on top of L2

L1 L2 close coupled

L2 L3 loose coupled 15mm

L3 L4 close coupled

L1 =impulse generation
L2=series resonant  (61nF)
L3= parallel resonant (35nF)
L4 =loaded (untuned)

Impulses do not show up, L2 and L3 are sine waves (l2 shows step up at peak like before).

F=81.13kC
input is 15W output is 7W (L4 rectified to DC into 15W resistance lamp) 147V dc 0.055A

I like the instant step up the L2 sine makes when L1 dumps it s energy into L2.

I made L2 and L3 sine waves equal size.

Without the lamp the resonance is undamped, and jumps up in voltage in both L2 and L3.
the power is higher without load
with load, the input power is lower.

no load: 
l2, l3=710V, 712V pp
1.6A 31.5V

15W load:
l2, l3= 247V,  243V pp
0.54A 31.5V

Both load and no load are resonant sine at max.

So far so good.

Now the new part...
L4 series resonant.

I'm using my rigol 2 channel square wave generator again, to drive the igbt half bridge.

I wonder if there is a smaller solution.
first thing that come to mind is the arduino due. it is fast enough to create 2 square waves oit of phase with 49% duty cycle.

maybe a prefab china thing?

I also checked signal between the 2 IGBT's. Doesnt look good. Ringing all over.

going to make gate negative first before checking into this. And that requires 2 new gate driver boards...

https://nl.mouser.com/datasheet/2/196/Infineon-IGW08T120-DS-v02_06-en-1225757.pdf

mouser link

found this IGBT
IKW08T120FKSA1
or IGW08T120
might be better. smaller amps, smaller charge
infineon trenchstop

maybe. maybe...
maybe I should make a tube half bridge

I keep playing with mosfet 2 coil configurations. to get both impulses in the series resonant coil.

I came up with this one.
L1 and L2 are both between halfbridge mosfets.

when both switches are conducting, coil charge (series resonant L2 and L1 impulse coil)

the mosfet open first alternating. this creates the impulses from L1. with alternating polarities.

the positve impulse always enters via the coil.
the negative impulse always enters via the cap.

the mosfet that is still conducting acts as ground. But that should not be a problem.

after the impulse event, both mosfets open for 50 or 48% duty cucle.
this is tricky.
as the series resonant coil now has no ground.
but it can discharge through L1.

worth testing. might be interesting.

Edit
if signals of fets are 75%duty and 180 degrees out of phase, there still will be a ground.

replaced igbt with mosfets again, and did test with previous setup.
it works half.

cap is on high side, and impulse shows up here.

but positive impulse does not show up from low side switch. Altough it is produced.

Will add series cap on low side of L2  also